we are going to neglect the pitch movement, the last DOF we need to discuss is the roll ϕ; we start from the computation of the lateral acceleration in the Roll Center, which is different from the one in the center of gravity in case the two centers do not correspond (i.e. if Hroll is not nil).
ay,RC = ay,G+ ¨ϕHrollcosϕ (2.37) from which we can compute the roll acceleration ¨ϕ
¨ϕ =
1may,RCHrollcosϕ2+1mgHrollsinϕ2−1Krollϕ+ G ˙ϕ2 Jx+ mHroll2
(2.38)
Figure 2.16: Map of the Sensors mounted on the EgoVehicle zoomed to see the relative positioning. Colour code comes from Simulink and is blue Cameras and red for Radars
2.6.1 Camera
The Camera was used for both Lane Detection and Object Detection, even though its main domain is the former, as the Object Detection (of the leading vehicle) is performed in cohabitation with the Central Radar of which we will discuss later.
For what concerns the Lane Detection, instead, the whole burden falls on the Camera [21] [33], as Radar is not able to distinguish colours and therefore it would not be capable of detecting the white lines delimiting the Lane, as well as geometric limitations meaning a Lidar installed in place of the Radar would only be able to recognize lane boundaries at some tens of meters from the Ego Vehicle. We are here listing the main parameters of the Camera, most of which were kept as default from the Simulink Vision Detection Generator block.
Figure 2.17: Map of the Sensors mounted on the EgoVehicle from a larger distance to see all the Sensors ranges
• Types of detections: Lanes and objects.
• Required interval between sensor updates: 0.1 s.
• Required interval between lane detection updates: 0.1 s.
• Sensor’s (x,y) position: [1.9 m, 0 m].
• Sensor’s height: 1.1 s.
• Yaw angle of sensor mounted: 0°.
• Pitch angle of sensor mounted: 1°.
• Roll angle of sensor mounted: 0°.
• Coordinate system used to report detections: Ego Cartesian.
• Maximum detection range: 30 m.
• Smoothing filter noise intensity: 5 m/s2.
• Maximum detectable object speed: 100 m/s.
• Maximum allowed occlusion for detector: 0.5.
• Minimum detectable image size of an object: [5 px, 5 px].
• Probability of detecting a target: 0.9.
• Number of false positives per image: 0.1.
• Minimum lane size in image: [20 px, 1 px].
• Accuracy of lane boundary 1 px.
• Focal length: [500 px, 800 px].
• Optical center of the camera: [320 px, 240 px].
• Yaw angle of sensor mounted: [480 px 640 px].
• Radial distortion coefficients: [0 0].
• Tangential distortion coefficients: [0 0].
• Skew of the camera axes: 0.
2.6.2 Central Radar
The Central Radar is tasked with the detection and tracking of an eventual leading vehicle and output the relative distance ∆X and the relative speed ∆V so that we can perform the Decision Making.
Most of the parameters were left as the default ones from the Simulink Radar Data Generator block, we are here reporting the most important ones.
• Update Rate: 10 Hz.
• Sensor mounting relative to Vehicle Origin: [1.9 m, 0 m, 0.2 m].
• Sensor rotation relative to Vehicle Frame: [0°, 0°, 0°].
• Azimuth resolution: 1°.
• Range resolution: 2.5 m.
• Range rate resolution: 0.5 m/s.
• Angular field of view [Azimuth, Elevation]: [20°, 5°].
• Range Limits: [0 m, 150 m].
• Range rate limits: [-100 m/s 100 m/s].
• Detection probability: 0.9.
• False alarm rate: 1e-06.
2.6.3 Side Radar
The Side Radar is tasked with the detection and tracking of oncoming vehicles and this guided us in its location and in the definition of its parameters.
In order to have a clear sight of the leftmost Lane, which is where we suppose to be detecting oncoming vehicles, we positioned this Radar on the front bumper of the EgoVehicle, on the leftmost part; meanwhile, we also rotated the Radar with respect to the non-inertial Reference Frame of the EgoVehicle, in order to have a detection area fully on the left of the vehicle.
• Update Rate: 10 Hz.
• Sensor mounting relative to Vehicle Origin: [3.7 m, 0.9 m, 0 m].
• Sensor rotation relative to Vehicle Frame: [7.5°, 0°, 0°].
• Azimuth resolution: 1°.
• Range resolution: 2.5 m.
• Range rate resolution: 0.5 m/s.
• Angular field of view [Azimuth, Elevation]: [15°, 5°].
• Range Limits: [0 m, 250 m].
• Range rate limits: [-100 m/s 100 m/s].
• Detection probability: 0.9.
• False alarm rate: 1e-06.
2.6.4 Blind Spot Radar
The Radar located near the Rear Right Wheel was dubbed "Blind Spot Radar"
because of its purpose, to detect any object in the rear right blindspot, which in Right-Hand Traffic countries (representing 65% of the world population and 70%
of the world roadways [76]) is where generally an overtaken vehicle is located. This is the actual purpose of such Radar, as will be discussed in Section 2.7.6. Below are reported the main parameters of the radar.
• Update Rate: 10 Hz.
• Sensor mounting relative to Vehicle Origin: [0 m, -0.9 m, 0.2 m].
• Sensor rotation relative to Vehicle Frame: [-105°, 0°, 0°].
• Azimuth resolution: 1°.
• Range resolution: 2.5 m.
• Range rate resolution: 0.5 m/s.
• Angular field of view [Azimuth, Elevation]: [150°, 5°].
• Range Limits: [0 m, 25 m].
• Range rate limits: [-100 m/s 100 m/s].
• Detection probability: 0.9.
• False alarm rate: 1e-06.
The main parameter is the very wide azimuth range, which was needed in order to be able to detect and track an overtaken vehicle. Since the overtaken vehicle will still be close to us, a long range was not needed, therefore we employed a short range of just 25 m, which is still sufficient to detect when we have completed the overtake. The relative yaw angle was chosen in order to have the boundary of the detection zone parallel to the longitudinal axis of the EgoVehicle.
2.6.5 Left BlindSpot Radar
As stated before, this Radar is only in a Beta-state, i.e. we have started some experimentation with it, but for reasons of simplicity and computational cost we have decided to leave it out for now. Its purpose was to detect if some vehicle behind us had already begun the Overtaking maneuver, meaning that we could not begin it ourselves in order to avoid an accident. Even if it was not part of our Sensor Configuration (hence why it is not present in Figures 2.16 and 2.17), we
decided to discuss it as well, since it can be used for further developments. Below are reported the main parameters of the radar.
• Update Rate: 10 Hz.
• Sensor mounting relative to Vehicle Origin: [0 m, 0.9 m, 0.2 m].
• Sensor rotation relative to Vehicle Frame: [135°, 0°, 0°].
• Azimuth resolution: 1°.
• Range resolution: 2.5 m.
• Range rate resolution: 0.5 m/s.
• Angular field of view [Azimuth, Elevation]: [90°, 5°].
• Range Limits: [0 m, 150 m].
• Range rate limits: [-100 m/s 100 m/s].
• Detection probability: 0.9.
• False alarm rate: 1e-06.